Mindaugas Žiūkas
Product Owner
Kaunas, Lithuania
Actions
I'm a developer and consultant at Softera Baltic, where I build and maintain a Business Central-based factoring solution for clients across Europe. I hold the MB-820 certification, and over the past year I've been using AI tools to speed up diagnosis across multi-layered AL code, spanning the product, client custom code, and BC standard.
I never expected factoring to be a domain that would grab me, but at Softera I found something unexpected - the people around me make it easy to care, whether you're building web pages or untangling complex business processes.
Links
Area of Expertise
Topics
Teaching AI a Vertical It Doesn't Know
Factoring — financing unpaid invoices — is one of the largest trade finance instruments globally yet almost unknown outside finance circles, including most BC developers who build vertical add-ons in domains they don't personally understand. This session spends 10 minutes making factoring concrete (what it actually is, why it isn't just "invoice lending"), then uses it as a case study for a broader question: how do you evaluate AI vendor claims for a domain you're not an expert in? Drawing from hands-on testing of AI document-processing vendors against real invoice data, and a live regulatory deadline (Poland's KSeF e-invoicing mandate) that is forcing AI adoption industry-wide right now, the talk gives developers a concrete framework: what to test, what AI reliably gets right in structured document extraction, and where it still needs a human who understands the domain's edge cases.
Debugging Across Three Layers: Using AI Coding Assistants to Triage Multi-Layered AL Extensions
Every BC ISV partner faces the same architecture problem: a bug report could originate in the core add-on, in a client's custom extension override, or in standard BC behavior the client misunderstood. Locating which layer actually owns the bug is the real bottleneck, not writing the fix. This session walks through a concrete triage method using AI coding assistants (Claude Code) to search and cross-reference product code, client custom code, and BC standard behavior before touching a single line, including a real case where this cut a feature investigation from an estimated multi-day effort to under a day. The talk covers what worked, what the AI got wrong, and where human judgment still has to override the tool.
Please note that Sessionize is not responsible for the accuracy or validity of the data provided by speakers. If you suspect this profile to be fake or spam, please let us know.
Jump to top